st: xi3 with char varname[omit] #

[Duru <duru80@gmail.com>]

Hi all,

I am a very fresh user, probably i will be bothering u for a simple
thing, but couldnt find it on the archives and it is urgent.

I will apply logistic regression in STATA to backcheck my "odds ratio"
calculations from the LEM output. Therefore, I need to use effect
coding and set the same reference categories as I used in my excel
calculations. The following lines work fine separately, but I couldn't
combine them.Other than gender and satisfaction variables, I will be
using the default first category- so I  am not mentioning them.

char GENDER[omit] 2
xi3 e.GENDER

char SATISFACTION[omit] 3
xi3 e.SATISFACTION

xi3 e.DISCRIMN  e.ETHNICITY  e.PERCENT  e.ETHNICITY*e.PERCENT

 Any tips on how to combine them before I run separate logistic for
the two effect coded categories of the dependent variable DISCRIMN? My
trials mostly gave the error message  "characteristic
SATISFACTION[omit] (3 GENDER[omit] 2) invalid;variable SATISFACTION is
numeric"


Thank in advance!



On Sun, Jun 6, 2010 at 7:13 PM, Nick Cox <n.j.cox@durham.ac.uk> wrote:
> As Martin said, -oversubopts()- is an undocumented option to my
> -catplot- (SSC). So, you'd need to look at the code to see that it was
> there -- or as Martin also implies look in the Statalist archives.
>
> It won't work anywhere else, unless exceptionally it's someone else's
> option name in their program.
>
> Back to the original question:
>
> 1. What is well documented is that -graph bar|hbar|dot- the y axis is
> whatever is the response axis, which need not be the vertical axis.
> -catplot- follows that principle exactly. Thus with -catplot hbar- the y
> axis is the horizontal axis.
>
> 2. Similarly, -catplot- like its -graph- parents regards the other axis
> as a categorical axis, and not as an x axis.
>
> 3. -oversubopts()- was an extra handle put into -catplot- for those who
> need it. Not documenting it was a good decision in that it appears that
> few users of -catplot- missed it and also said so in my sight; and a bad
> decision in so far various handles for tweaking the display were closed
> off unless you looked.
>
> !!! note the caveat below
>
> In my files, but not yet public, is a revised version of -catplot- in
> which the command in question would be
>
> catplot rep78 foreign, percent(foreign) stack asyvars legend(span
> rows(1)) var2opts(label(labsize(small)))
>
> Here -var2opts()- is also idiosyncratic to -catplot-, but it is
> documented in the help for -catplot-.
>
> I say again: this syntax is not accessible yet except to myself and a
> couple of kind guinea-pigs (otherwise known as "beta testers").
>
> Nick
> n.j.cox@durham.ac.uk
>
> Kaulisch, Marc
>
> I got lost with this sentence in h graph hbar
> "The over_subopts -- used in over(varname, over_subopts) and, on rare
> occasion, in
>    yvaroptions(over_subopts) -- are"
>
> I haven't interpreted it as oversubopts can stand alone. My Stata
> experience is not so good to grap the sense of a code in a minute...
>
> Martin Weiss
>
> You need a certain level of experience to tell that there must be more
> to the command than meets the eye in its help file. The sentence in -h
> catplot- "graph_options refers to options of graph bar, graph hbar or
> graph dot as appropriate.  by() is one useful example." tells you that
> it is worth investigating. At the same time, you know that Nick`s
> -program- is a derivative of an existing -graph- command, so you can
> -trace- the thing and see that it issues a call to a certain official
> command. Once you know that, you look at the code itself and determine
> how you can add options to that call...
>
> Kaulisch, Marc
>
> Thanks a lot. Somehow I haven't seen the tree in the forest... I haven't
> thought that oversubopts can be used as an option on its own...
>
> Martin Weiss
>
> Marc may also want to take a look at NJC`s statement here:
> http://www.stata.com/statalist/archive/2009-07/msg00974.html
>
> Martin Weiss
>
> You can pass -undocumented- "oversubopts" to -ssc d catplot-, though:
>
> ***********
> sysuse auto, clear
> catplot hbar rep78 foreign, percent(foreign) ///  stack asyvars
> ylabel(,labsize(vsmall)) ///  legend(span rows(1))
> oversubopts(label(labsize(small)))
> ***********
>
> Kaulisch, Marc
>
> I might be again missing some features or am not carefully enough
> reading the help files, but I have problems to control labels of the
> y-axis when I use catplot (ssc) (I would like to avoid to use the graph
> editor...). I tried several options but I cannot get a solution (besides
> using the graph editor).
>
> My code looks like:
> sysuse auto
> catplot hbar rep78 foreign, percent(foreign) stack asyvars
> ylabel(,labsize(vsmall)) legend(span rows(1))
>
> ylabel controls the x-axis labels; but if xlabel spcified:
> xlabels(,labsize(vsmall)) not allowed, xaxis1 does not exist
>
> The yvar-option does not work either:
> catplot hbar rep78 foreign, percent(foreign) stack asyvars
> ylabel(,labsize(vsmall)) legend(span rows(1)) yvar(label(labs(vsmall)))
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/